NOVBMBEE 16, 1900.] 



SCIENCE. 



741 



a compliancy, on increasing H, fj. will de- 

 crease by an amount depending upon the 

 first power of H, and on increasing F, k 

 will decrease by an amount depending 

 upon the second power of F. Also, in this 

 latter case, the diminution of k must depend 

 inversely upon the coefficient of volume 

 elasticity. 



On the other hand, if it is k which is the 

 compliancy, these relatioas will be inter- 

 changed. 



It was at once noticed that several of the 

 empirical formulae expressing the relation 

 between S and /i gave a diminution de- 

 pending upon the first power of H. A 

 somewhat elaborate investigation was then 

 undertaken, extending over the greater part 

 of a year, and the fact was definitely estab- 

 lished that the diminution did depend upon 

 the first power of inaccurately, the maxi- 

 mum amount of deviation from that called 

 for being less than one-fourth of one per 

 cent., which was about the limit of experi- 

 mental accuracy. 



This, of itself, would have sufficed to 

 have settled the point, but in addition the 

 other relation, which should exist if the 

 theory were correct, i. e., that the specific 

 inductive capacity, k, should vary with the 

 inverse second power of the slope of electric 

 potential, and as the coefficient of volume 

 elasticity, was also discovered. This was 

 found to be the complete expression of 

 Kerr's electrostatic phenomenon. 



A prism of glass, one cm. thick and one 

 cm. wide, stretched with a force of 30.10° 

 dynes gave a change of density of nearly 

 3.10 "" ^ The change in the thickness of the 

 glass was approximately 1.5.10~°. The 

 change in velocity of the light which passed 

 transversely through the glass, was approxi- 

 mately .7 X 10 -\ 



It was thus found that the actual me- 

 chanically produced change in density of 

 the glass was sufiicient to account for the 

 observed change of velocity, though the 



agreement was not so close as it might have 

 been, possibly owing to experimental diffi- 

 culties. 



From the observed change in velocity 

 when placed in a strong electrostatic field, 

 whose value was approximately determined 

 by its sparkling distance, it was calculated 

 that the value of the F'^jSnk stress required 

 to produce the same change of velocitj' as 

 had been produced mechanically was nearly 

 25.10° dynes. The value of the purely me- 

 chanically applied stress, as given above, 

 was 30.10°. The close agreement is prob- 

 ably accidental, as the experimental error 

 was considerably greater than the small dif- 

 ference observed. It is intended to repeat 

 these experiments under conditions permit-- 

 ting of a much higher degree of accuracy. 



The results obtained are however suffi- 

 cient to show that Kerr's effect can be ac- 

 counted for by purely mechanical stresses, 

 electrically produced and resulting in a 

 change of density. 



ISTow it was pointed out above, that which- 

 ever of the medium coefficients, k or //, va- 

 ries as the square of the corresponding in- 

 tensity, that one must be a density. Since, 

 therefore, it has already been shown by 

 Kerr that the change in velocity, and hence, 

 as my experiments prove, the change in 

 density,* is proportional to the square of the 

 electric intensity, it follows that i is a den- 

 sity. 



It still remained to be shown that Kerr'B 

 efifect depended upon the volume elasticity. 

 This was done by testing different glasses 

 and noting that, the compensating pieces 

 being made from the glass under test, the 

 same force was always required to compen- 

 sate, independent of the material tested. 



We see, therefore, that the results de- 

 duced from the experiments on the relation 

 between if and ij. are completely confirmed 



* Velocity is proportional to square root of density, 

 but change of velocity is proportional to change in 

 density, both being small. 



